Rate limiting-Alibaba Cloud Model Studio(Model Studio)-阿里云帮助中心

Model Studio rate limiting rules and the rate limits of each model. Major text generation models use dynamic rate limiting (soft rate limiting): the TPM rate limit value is adjusted monthly based on your Model Studio monthly consumption tier, and the actual available TPM is no lower than the rate limit value. Other models and regions use static rate limiting: requests are rejected when the limit is exceeded and typically recover automatically within one minute. Rate limits are applied at the account level, aggregating usage across RAM users, workspaces, and API keys.

Rate limiting rules

Account-level rate limiting: Rate limits are applied at the root account level. The usage of all RAM users, workspaces, and API keys under the account is combined.
Model-specific rate limiting: Each model has its own rate limit. For more information, see the tables below.
Rate limiting types: Major text generation models (such as qwen3.8-max, qwen3.8-flash, deepseek-v4-pro-0813, and kimi-k3) use dynamic rate limiting, where the TPM rate limit value is adjusted monthly based on your Model Studio monthly consumption tier. For more information, see Dynamic rate limiting. Other models and regions use static rate limiting; see Static rate limiting rules.

Dynamic rate limiting

Dynamic rate limiting mechanism

Bailian applies a dynamic rate limiting mechanism to some models. The key points are as follows:

Rate limit value: Determined by tier based on your monthly consumption amount on Bailian. It serves as the guaranteed baseline for throughput and is adjusted every month.
Soft rate limiting: When the call volume reaches the rate limit value, if the platform still has spare resources, rate limiting is not triggered. Therefore, the actual available TPM is no lower than the rate limit value. The actual rate limit value observed during calls may be higher than this rate limit value, which is expected behavior.
Rate limit granularity: By account + model. The call volumes of all RAM users, workspaces, and API Keys under a primary account are aggregated and counted toward the TPM quota for that account for that model, sharing the same limit.
Workspace: You can set a separate rate limit value for a model in a workspace. Workspaces without a separate setting share the account-level quota.
Behavior when rate limited: When dynamic rate limiting is triggered, a 429 error is returned (error code Throttling.ResourceExhausted or Throttling.ServiceOverloaded, with the message Too many requests.). If the platform is temporarily short of resources, a 503 error (ServiceUnavailable) may be returned, which is unrelated to the rate limit quota — retry later. For error handling, see Error codes.
More information: For workspace rate limit settings, see Permission management.

Models and rate limit tiers

The TPM rate limit value is tiered based on the monthly consumption amount on the Bailian platform. The tiers for supported models in different regions are as follows (values are subject to the actual display in the console):

China (Beijing)

Bailian platform monthly consumption:	Tier 1 (current) ≤¥100,000	Tier 2 ¥100,000~¥1,000,000	Tier 3 >¥1,000,000
Supported models	TPM rate limit tiers
qwen3.8-max	5,000,000	10,000,000	20,000,000
qwen3.8-max-0902	1,500,000	1,500,000	1,500,000
qwen3.8-flash	5,000,000	10,000,000	20,000,000
qwen3.8-omni-flash	2,000,000	5,000,000	10,000,000
qwen3.8-2.4t-a95b	5,000,000	5,000,000	5,000,000
deepseek-v4.1-flash-1017	1,200,000	1,200,000	1,200,000
deepseek-v4.1-flash	1,200,000	1,200,000	1,200,000
deepseek-v4-pro-0813	1,200,000	10,000,000	20,000,000
kimi-k3 kimi-k3.1-pro	1,200,000	5,000,000	10,000,000
glm-5.3	5,000,000	5,000,000	5,000,000
glm-5.3-prime	2,000,000	2,000,000	2,000,000
qwen3.8-27b	5,000,000	5,000,000	5,000,000

Singapore

Bailian platform monthly consumption:	Tier 1 (current) ≤¥100,000	Tier 2 ¥100,000~¥1,000,000	Tier 3 >¥1,000,000
Supported models	TPM rate limit tiers
qwen3.8-max	2,000,000	5,000,000	10,000,000
qwen3.8-max-0902	1,500,000	1,500,000	1,500,000
qwen3.8-flash	2,500,000	5,000,000	10,000,000
qwen3.8-omni-flash	2,000,000	5,000,000	10,000,000
qwen3.8-2.4t-a95b	5,000,000	5,000,000	5,000,000
deepseek-v4.1-flash-1017	1,200,000	1,200,000	1,200,000
deepseek-v4.1-flash	1,200,000	1,200,000	1,200,000
deepseek-v4-pro-0813	1,200,000	5,000,000	10,000,000
kimi-k3 kimi-k3.1-pro	1,200,000	5,000,000	10,000,000
glm-5.3	5,000,000	5,000,000	5,000,000
glm-5.3-prime	2,000,000	2,000,000	2,000,000
qwen3.8-27b	5,000,000	5,000,000	5,000,000

Take China (Beijing) as an example: if the Bailian consumption amount of an account in the previous calendar month was 500,000, it falls into the Tier 2 (¥100,000~¥1,000,000) tier, corresponding to a TPM rate limit value of 10,000,000 for qwen3.8-max.

Quota display

Log in to the Bailian console. In the left-side navigation pane, choose Model rate limits to go to the model rate limits page.

The model rate limits page is a read-only page that displays the models that support dynamic rate limiting, the current TPM rate limit value, and the Bailian consumption tier. The top of the page shows the current effective period (such as 2026-08-15 to 2026-09-15), last month's consumption (calculated by the previous calendar month, such as August 1 to August 31), and this month's consumption amount. You can click View consumption details to view consumption details, and click View rules to view the rate limit tier rules. The table below lists the TPM rate limit tiers for each model.

NoteIf the TPM rate limit value cannot meet your business needs, contact your account manager to apply for a quota increase.

Dynamic rate limit adjustment strategy

Rate limit tiers are dynamically adjusted according to the following rules:

Tiering basis: Bailian platform consumption amount.
Calculation cycle: Calculated by calendar-month billing cycle.
Notification: A notification is sent to the affected users on the 10th of each month.
Effective time: The new tier takes effect on the 15th of each month and is valid until the 15th of the following month.
Adjustment direction: Lower tier → higher tier (upgrade) or remain unchanged.

Static rate limits

The rate limits of each model are listed by model category below: models marked dynamic rate limiting have TPM rate limit values adjusted by monthly consumption tier. For more information, see Models and rate limit tiers. All other models use static rate limiting; see Static rate limiting rules.

Static rate limiting rules

Models and regions not covered by dynamic rate limiting use static rate limiting:

Fixed rate limits: Each model has fixed RPM and TPM rate limits, independent of other models.
Per-second enforcement: In addition to per-minute RPM/TPM, rate limiting can also be enforced per second as RPS (RPM/60) and TPS (TPM/60). A burst of requests in a short period can trigger rate limiting even if the per-minute limits are not exceeded.
Behavior when triggered: Requests that exceed the limits are rejected with a 429 error and typically recover automatically within one minute. For error message classification, see Why is rate limiting triggered? and Error codes.

Text generation

Qwen language model

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-max Rate limiting does not apply when calling the service with the Batch API.	Dynamic rate limiting
qwen3.8-max-0902	Dynamic rate limiting
qwen3.8-flash Rate limiting does not apply when calling the service with the Batch API.	Dynamic rate limiting
qwen3.7-max Rate limiting does not apply when calling the service with the Batch API.	30,000	5,000,000
qwen3.7-max-2026-06-08	600	1,000,000
qwen3.7-max-2026-05-20	600	1,000,000
qwen3.7-max-preview	60	500,000
qwen3.7-max-2026-05-17	60	500,000
qwen-max Rate limiting does not apply when calling the service with the Batch API.	1,200	1,000,000
qwen3.7-plus	30,000	5,000,000
qwen3.7-plus-2026-05-26	600	1,000,000
qwen3.6-plus Rate limiting does not apply when calling the service with the Batch API.	30,000	5,000,000
qwen3.6-plus-2026-04-02	600	1,000,000
qwen3.7-flash Rate limiting does not apply when calling the service with the Batch API.	30,000	5,000,000
qwen3.7-flash-2026-07-15	30,000	5,000,000
qwen3.6-flash Rate limiting does not apply when calling the service with the Batch API.	30,000	10,000,000
qwen3.6-flash-2026-04-16	600	1,000,000
qwen3.5-plus Rate limiting does not apply when calling the service with the Batch API.	30,000	5,000,000
qwen3.5-plus-2026-04-20	600	1,000,000
qwen3.5-plus-2026-02-15	600	1,000,000
qwen-plus Rate limiting does not apply when calling the service with the Batch API.	30,000	5,000,000
qwen-plus-latest Rate limiting does not apply when calling the service with the Batch API.	15,000	1,200,000
qwen-plus-2025-12-01	120	1,000,000
qwen-plus-2025-07-14 (qwen-plus-0714)	60	100,000
qwen-plus-2025-04-28 (qwen-plus-0428)	60	1,000,000
qwen-plus-2025-01-25 (qwen-plus-0125)	60	150,000
qwen3.5-flash Rate limiting does not apply when calling the service with the Batch API.	30,000	10,000,000
qwen3.5-flash-2026-02-23	600	1,000,000
qwen-flash Rate limiting does not apply when calling the service with the Batch API.	30,000	10,000,000
qwen-flash-2025-07-28	60	1,000,000
qwen-long Rate limiting does not apply when calling the service with the Batch API.	1,200	3,000,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-max	Global	30,000	5,000,000
qwen3.8-max	US	30,000	5,000,000
qwen3.8-max-0902	Global	30,000	150,000
qwen3.8-flash	Global	30,000	5,000,000
qwen3.8-flash	US	15,000	2,500,000
qwen3.7-max	Global	30,000	5,000,000
qwen3.7-max	US	600	1,000,000
qwen3.7-max-2026-06-08	Global	600	1,000,000
qwen3.7-max-2026-05-20	Global	600	1,000,000
qwen3.7-plus	Global	30,000	5,000,000
qwen3.7-plus	US	15,000	5,000,000
qwen3.7-plus-2026-05-26	Global	600	1,000,000
qwen3.6-plus	Global	30,000	5,000,000
qwen3.6-plus-2026-04-02	Global	600	1,000,000
qwen3.7-flash	Global	15,000	5,000,000
qwen3.7-flash	US	15,000	5,000,000
qwen3.7-flash-2026-07-15	Global	60	1,000,000
qwen3.6-flash	Global	15,000	5,000,000
qwen3.6-flash-2026-04-16	Global	60	1,000,000
qwen3.6-flash	US	15,000	5,000,000
qwen3.5-plus	Global	30,000	5,000,000
qwen3.5-plus-2026-02-15	Global	600	1,000,000
qwen-plus	Global	15,000	5,000,000
qwen-plus	US	600	1,000,000
qwen-plus-2025-12-01	Global	60	1,000,000
qwen-plus-2025-12-01	US	60	1,000,000
qwen3.5-flash	Global	30,000	10,000,000
qwen3.5-flash-2026-02-23	Global	600	1,000,000
qwen-flash	Global	15,000	10,000,000
qwen-flash	US	30000	10,000,000
qwen-flash-2025-07-28	Global	60	1,000,000
qwen-flash-2025-07-28	US	600	5,000,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-max	International	Dynamic rate limiting
qwen3.8-max-0902	International	Dynamic rate limiting
qwen3.8-flash	International	Dynamic rate limiting
qwen3.7-max	International	600	1,000,000
qwen3.7-max-2026-06-08	International	60	1,000,000
qwen3.7-max-2026-05-20	International	60	1,000,000
qwen-max Rate limiting does not apply when calling the service with the Batch API.	International	600	1,000,000
qwen3.7-plus	International	15,000	5,000,000
qwen3.7-plus-2026-05-26	International	60	1,000,000
qwen3.6-plus	International	15,000	5,000,000
qwen3.6-plus-2026-04-02	International	60	1,000,000
qwen3.7-flash	International	15,000	5,000,000
qwen3.7-flash-2026-07-15	International	15,000	5,000,000
qwen3.6-flash	International	15,000	5,000,000
qwen3.6-flash-2026-04-16	International	60	1,000,000
qwen3.5-plus	International	15,000	5,000,000
qwen3.5-plus-2026-04-20	International	600	1,000,000
qwen3.5-plus-2026-02-15	International	60	1,000,000
qwen-plus-latest	International	600	1,000,000
qwen-plus-2025-12-01	International	120	1,000,000
qwen-plus-2025-07-14 (qwen-plus-0714)	International	60	100,000
qwen-plus-2025-04-28 (qwen-plus-0428)	International	60	1,000,000
qwen-plus-2025-01-25 (qwen-plus-0125)	International	60	100,000
qwen3.5-flash	International	15,000	5,000,000
qwen3.5-flash-2026-02-23	International	60	1,000,000
qwen-flash	International	600	5,000,000
qwen-flash-2025-07-28	International	600	5,000,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-max	Global	30,000	5,000,000
qwen3.8-max-0902	Global	30,000	150,000
qwen3.8-flash	EU	30,000	5,000,000
qwen3.8-flash	Global	30,000	5,000,000
qwen3.7-max	Global	30,000	5,000,000
qwen3.7-max-2026-06-08	Global	600	1,000,000
qwen3.7-max-2026-05-20	Global	600	1,000,000
qwen3.7-plus	Global	30,000	5,000,000
qwen3.7-plus-2026-05-26	Global	600	1,000,000
qwen3.6-plus	Global	30,000	5,000,000
qwen3.6-plus-2026-04-02	Global	600	1,000,000
qwen3.7-flash	Global	15,000	5,000,000
qwen3.7-flash-2026-07-15	Global	15,000	5,000,000
qwen3.6-flash	Global	15,000	5,000,000
qwen3.6-flash-2026-04-16	Global	60	1,000,000
qwen3.5-plus	Global	30,000	5,000,000
qwen3.5-plus-2026-02-15	Global	600	1,000,000
qwen-plus	Global	600	1,000,000
qwen-plus	EU	600	1,000,000
qwen-plus-2025-12-01	Global	120	1,000,000
qwen-plus-2025-12-01	EU	120	1,000,000
qwen3.5-flash	Global	35,000	10,000,000
qwen3.5-flash	EU	35,000	10,000,000
qwen3.5-flash-2026-02-23	Global	600	1,000,000
qwen3.5-flash-2026-02-23	EU	600	1,000,000
qwen-flash	Global	30,000	10,000,000
qwen-flash-2025-07-28	Global	60	1,000,000

Japan (Tokyo)

Model name	Service deployment scope	Rate limit conditions (rate limiting is triggered when any value is exceeded) The following are per-minute rate limits. The service may also enforce per-second limits at RPS (RPM/60) and TPS (TPM/60)
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Including input and output tokens
qwen3.8-max	Global	30,000	5,000,000
qwen3.8-max-0902	Global	30,000	150,000
qwen3.8-flash	Global	30,000	5,000,000
qwen3.7-max	Global	600	1,000,000
qwen3.7-max-2026-05-20	Global	60	1,000,000
qwen3.7-plus	Global	15,000	5,000,000
qwen3.7-plus-2026-05-26	Global	60	1,000,000
qwen3.7-plus	Japan	15,000	5,000,000
qwen3.7-plus-2026-05-26	Japan	60	1,000,000
qwen3.6-plus	Global	15,000	5,000,000
qwen3.6-plus-2026-04-02	Global	60	1,000,000
qwen3.7-flash	Global	15,000	5,000,000
qwen3.7-flash-2026-07-15	Global	15,000	5,000,000
qwen3.6-flash	Global	15,000	10,000,000
qwen3.6-flash-2026-04-16	Global	60	1,000,000

Hong Kong (China)

Model name	Service deployment scope	Rate limiting conditions (triggered if any value is exceeded) The following limits are per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens.
qwen3.8-max	Hong Kong (China)	600	1,000,000
qwen3.8-max-0902	Hong Kong (China)	600	150,000
qwen3.8-flash	Hong Kong	30,000	5,000,000
qwen3.8-flash	Global	30,000	5,000,000
qwen3.6-plus	Global	30,000	5,000,000
qwen3.7-flash	Global	15,000	5,000,000
qwen3.7-flash-2026-07-15	Global	15,000	5,000,000
qwen3.6-flash	Global	30,000	10,000,000
qwen-plus	Hong Kong (China)	600	1,000,000
qwen-plus-2025-12-01	Hong Kong (China)	60	100,000
qwen3.5-flash	Hong Kong (China)	30,000	10,000,000
qwen3.5-flash-2026-02-23	Hong Kong (China)	600	1,000,000

Qwen-VL (visual understanding/image-to-text)

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
qwen3-vl-plus Rate limiting does not apply when calling the service with the Batch API.	3,000	5,000,000

Model name

Rate limits (triggered if any value is exceeded)

The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).

Requests per minute (RPM)

Tokens per minute (TPM)

Includes input and output tokens

qwen3-vl-plus

Rate limiting does not apply when calling the service with the Batch API.

3,000

5,000,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3-vl-plus	Global	6,000	10,000,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3-vl-plus	International	1,200	1,000,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3-vl-plus	Global	2,000	1,000,000
qwen3-vl-plus	EU	2,000	1,000,000

Hong Kong (China)

Model name	Service deployment scope	Rate limiting conditions (triggered if any value is exceeded) The following limits are per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens.
qwen3-vl-plus	Hong Kong (China)	6,000	1,000,000

Qwen-Omni

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-omni-flash	30000	Dynamic rate limits
qwen3.5-omni-plus	60	100,000
qwen3.5-omni-plus-2026-03-15	60	100,000
qwen3.5-omni-flash	60	100,000
qwen3.5-omni-flash-2026-03-15	60	100,000
qwen3-omni-flash	60	100,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-omni-flash	International	30000	Dynamic rate limits
qwen3.5-omni-plus	International	60	100,000
qwen3.5-omni-plus-2026-03-15	International	60	100,000
qwen3.5-omni-flash	International	60	100,000
qwen3.5-omni-flash-2026-03-15	International	60	100,000
qwen3-omni-flash	International	60	100,000

Hong Kong (China)

Model name	Service deployment scope	Rate limiting conditions (triggered if any value is exceeded) The following limits are per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens.
qwen3.8-omni-flash	Global	30000	2000000

Japan (Tokyo)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-omni-flash	Global	30000	2000000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-omni-flash	Global	30000	2000000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-omni-flash	Global	30000	2000000

Qwen-Omni-Realtime

For WebSocket connections, RPM limits the number of connection requests per minute. Responses on an existing connection do not count toward RPM. Reconnecting counts as a new request. There is currently no separate limit on the total number of connections open at the same time. TPM limits still apply. See the tables below for model limits.

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-omni-flash-realtime	60	2,000,000
qwen3.5-omni-plus-realtime	60	100,000
qwen3.5-omni-plus-realtime-2026-03-15	60	100,000
qwen3.5-omni-flash-realtime	60	100,000
qwen3.5-omni-flash-realtime-2026-03-15	60	100,000
qwen3-omni-flash-realtime-2025-09-15	60	100,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-omni-flash-realtime	International	60	2,000,000
qwen3.5-omni-plus-realtime	International	60	100,000
qwen3.5-omni-plus-realtime-2026-03-15	International	60	100,000
qwen3.5-omni-flash-realtime	International	60	100,000
qwen3.5-omni-flash-realtime-2026-03-15	International	60	100,000
qwen3-omni-flash-realtime-2025-09-15	International	60	100,000

Qwen-OCR (text extraction)

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.5-ocr	6,000	30,000,000
qwen-vl-ocr Rate limiting does not apply when calling the service with the Batch API.	600	6,000,000
qwen-vl-ocr-2025-11-20	6,000	30,000,000
qwen-vl-ocr-2024-10-28	600	6,000,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-vl-ocr	Global	600	6,000,000
qwen-vl-ocr-2025-11-20	Global	1,200	6,000,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-vl-ocr	International	600	6,000,000
qwen-vl-ocr-2025-11-20	International	1,200	6,000,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-vl-ocr	Global	600	6,000,000
qwen-vl-ocr-2025-11-20	Global	1,200	6,000,000

Qwen Audio (audio understanding)

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-audio-turbo	120	100,000
qwen-audio-turbo-latest	60

Qwen-Coder

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3-coder-flash	5,000	5,000,000
qwen3-coder-flash-2025-07-28	60	1,000,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3-coder-flash	Global	1,200	1,000,000
qwen3-coder-flash-2025-07-28	Global	60	1,000,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3-coder-flash	International	600	5,000,000
qwen3-coder-flash-2025-07-28	International	600	5,000,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3-coder-flash	Global	5,000	5,000,000
qwen3-coder-flash-2025-07-28	Global	60	1,000,000

Qwen translation model

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-mt-plus	60	25,000
qwen-mt-flash	60	35,000
qwen-mt-lite	60	100,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-mt-plus	Global	60	25,000
qwen-mt-flash	Global	60	35,000
qwen-mt-lite	Global	60	100,000
qwen-mt-lite	US	60	100,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-mt-plus	International	60	100,000
qwen-mt-flash	International	60	100,000
qwen-mt-lite	International	60	100,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-mt-plus	Global	60	25,000
qwen-mt-flash	Global	60	35,000
qwen-mt-lite	Global	60	100,000

Qwen-MT-Uni

China (Beijing)

Model name	Rate limit (triggered when any threshold is exceeded)
Model name	Requests per minute (RPM)	Tokens per minute (TPM)	Async concurrency	Async queue size
qwen-mt-uni	162	2,592,000	3	500

Qwen data mining model

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-doc-turbo	600	3,000,000

Qianwen In-depth Research Model

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-deep-research	120	1,200,000

Tongyi Xiaomì conversation analysis model

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
tongyi-xiaomi-analysis-flash	600	1,000,000
tongyi-xiaomi-analysis-pro	600	1,000,000

Qwen language model open source version

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-2.4t-a95b	Dynamic rate limiting
qwen3.8-27b	Dynamic rate limiting
qwen3.6-35b-a3b	600	1,000,000
qwen3.6-27b	600	1,000,000
qwen3.5-397b-a17b	600	1,000,000
qwen3.5-122b-a10b	600	1,000,000
qwen3.5-27b	600	1,000,000
qwen3.5-35b-a3b	600	1,000,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.6-35b-a3b	Global	600	1,000,000
qwen3.5-397b-a17b	Global	600	1,000,000
qwen3.5-122b-a10b	Global	600	1,000,000
qwen3.5-27b	Global	600	1,000,000
qwen3.5-35b-a3b	Global	600	1,000,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.8-2.4t-a95b	International	Dynamic rate limiting
qwen3.8-27b	International	Dynamic rate limiting
qwen3.6-35b-a3b	International	600	1,000,000
qwen3.6-27b	International	600	1,000,000
qwen3.5-397b-a17b	International	600	1,000,000
qwen3.5-122b-a10b	International	600	1,000,000
qwen3.5-27b	International	600	1,000,000
qwen3.5-35b-a3b	International	600	5,000,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.6-35b-a3b	Global	600	1,000,000
qwen3.5-397b-a17b	Global	600	1,000,000
qwen3.5-122b-a10b	Global	600	1,000,000
qwen3.5-27b	Global	600	1,000,000
qwen3.5-35b-a3b	Global	600	1,000,000

DeepSeek

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
deepseek-v4.1-flash-1017	Dynamic rate limiting
deepseek-v4.1-flash	Dynamic rate limiting
deepseek-v4-pro	15,000	1,200,000
deepseek-v4-pro-0813	Dynamic rate limiting
deepseek-v4-flash-0731	15,000	1,200,000
deepseek-r1-distill-qwen-1.5b	60	100,000
deepseek-r1-distill-llama-70b	60	100,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
deepseek-v4.1-flash-1017	Global	15,000	1,200,000
deepseek-v4.1-flash	Global	15,000	1,200,000
deepseek-v4-pro	Global	15,000	1,200,000
deepseek-v4-pro-0813	Global	15,000	1,200,000
deepseek-v4-pro	US	10,000	1,200,000
deepseek-v4-flash-0731	Global	15,000	1,200,000
deepseek-v4-flash-0731	US	15,000	1,200,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
deepseek-v4.1-flash-1017	International	Dynamic rate limiting
deepseek-v4.1-flash	International	Dynamic rate limiting
deepseek-v4-pro	International	10,000	1,200,000
deepseek-v4-pro-0813	International	Dynamic rate limiting
deepseek-v4-flash-0731	International	15,000	1,200,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
deepseek-v4.1-flash-1017	Global	15,000	1,200,000
deepseek-v4.1-flash	Global	15,000	1,200,000
deepseek-v4-pro	Global	15,000	1,200,000
deepseek-v4-pro-0813	Global	15,000	1,200,000
deepseek-v4-flash-0731	Global	15,000	1,200,000

Japan (Tokyo)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
deepseek-v4-pro	Japan	10,000	1,200,000
deepseek-v4-pro-0813	Japan	15,000	1,200,000
deepseek-v4-flash-0731	Japan	15,000	1,200,000
deepseek-v4.1-flash-1017	Global	15,000	1,200,000
deepseek-v4.1-flash	Global	15,000	1,200,000
deepseek-v4-pro	Global	10,000	1,200,000
deepseek-v4-pro-0813	Global	15,000	1,200,000
deepseek-v4-flash-0731	Global	15,000	1,200,000

Hong Kong (China)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
deepseek-v4.1-flash-1017	Hong Kong (China)	15,000	1,200,000
deepseek-v4.1-flash	Hong Kong (China)	15,000	1,200,000
deepseek-v4-flash-0731	Hong Kong (China)	15,000	1,200,000
deepseek-v4-pro-0813	Hong Kong (China)	15,000	1,200,000

DeepSeek-SiliconFlow direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
siliconflow/deepseek-v3.2	500	500,000
siliconflow/deepseek-v3.1-terminus	500	500,000
siliconflow/deepseek-r1-0528	500	500,000
siliconflow/deepseek-v3-0324	500	500,000

DeepSeek-Kuaishou Vanchin direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
vanchin/deepseek-v3.2-think	30	600,000
vanchin/deepseek-v3.1-terminus	500	1,000,000
vanchin/deepseek-r1	500	1,000,000
vanchin/deepseek-v3	500	1,000,000
vanchin/deepseek-ocr	500	1,000,000
vanchin/deepseek-v4-pro	60	300,000
vanchin/deepseek-v4-pro-0813	30	600,000
vanchin/deepseek-v4.1-flash	30	600,000

Kimi

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
kimi-k3.1-pro	15,000	Dynamic rate limiting
kimi-k3	Dynamic rate limiting
kimi-k2.7-code	500	1,000,000
kimi-k2.6	500	1,000,000
kimi-k2.5	500	1,000,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
kimi-k3 kimi-k3.1-pro	Global	15,000	1,200,000
kimi-k3.1-pro	US	Rate limits depend on the actual configuration of the selected workspace.
kimi-k3	International	10,000	1,200,000
kimi-k2.7-code	Global	500	1,000,000
kimi-k2.5	Global	500	1,000,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
kimi-k3 kimi-k3.1-pro	Global	15,000	1,200,000
kimi-k2.7-code	Global	500	1,000,000
kimi-k2.5	Global	500	1,000,000

Japan (Tokyo)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
kimi-k3 kimi-k3.1-pro	Global	15,000	1,200,000
kimi-k2.7-code	Global	500	1,000,000
kimi-k2.5	Global	500	1,000,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
kimi-k3.1-pro	International	10,000	Dynamic rate limiting
kimi-k3	International	Dynamic rate limiting
kimi-k2.7-code	International	500	1,000,000

Hong Kong (China)

Model name

Service deployment scope

Rate limits (triggered if any value is exceeded)

The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).

Requests per minute (RPM)

Tokens per minute (TPM)

Includes input and output tokens

kimi-k3

kimi-k3.1-pro

Global

15,000

1,200,000

kimi-k2.7-code

Global

500

1,000,000

Kimi-Moonshot AI direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
kimi/kimi-k3.1-pro-highspeed	500 Under the same Model Studio API key, a shared rate limit of 500 RPM is applied across the seven models. This means the total number of requests per minute for these seven models cannot exceed 500.	3,000,000 Under the same Model Studio API key, a shared rate limit of 3,000,000 TPM is applied across the seven models. This means the total number of tokens consumed per minute by these models cannot exceed 3,000,000.
kimi/kimi-k3.1-pro
kimi/kimi-k3
kimi/kimi-k2.7-code-highspeed
kimi/kimi-k2.7-code
kimi/kimi-k2.6
kimi/kimi-k2.5

GLM

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
glm-5.3	Dynamic rate limiting
glm-5.2	500	2,000,000
glm-5.1	500	1,000,000
glm-5	500	1,000,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
glm-5.3	Global	15,000	1,200,000
glm-5.3	US	15,000	1,200,000
glm-5.2	Global	500	1,000,000
glm-5.2	US	500	1,000,000
glm-5.1	Global	500	1,000,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
glm-5.3	Global	15,000	1,200,000
glm-5.2	Global	500	1,000,000
glm-5.1	Global	500	1,000,000

Japan (Tokyo)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
glm-5.3	Global	15,000	1,200,000
glm-5.1	Global	500	1,000,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
glm-5.3	International	Dynamic rate limiting
glm-5.2	International	500	1,000,000
glm-5.1	International	500	1,000,000

Hong Kong (China)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
glm-5.2	Global	500	1,000,000

GLM-Zhipu AI direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
ZHIPU/GLM-5.3	200	3,000,000
ZHIPU/GLM-5.3-Flash	200	3,000,000
ZHIPU/GLM-5.3-FlashX	200	3,000,000
ZHIPU/GLM-5.2	200	3,000,000
ZHIPU/GLM-5.1	200	3,000,000
ZHIPU/GLM-5	200	3,000,000

MiniMax

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
MiniMax-M2.5	500	1,000,000

MiniMax-Xiyu Tech direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
MiniMax/MiniMax-M3.1-Flash	200	10,000,000
MiniMax/MiniMax-M3	500	20,000,000
MiniMax/MiniMax-M2.7	500	20,000,000
MiniMax/MiniMax-M2.5	500	20,000,000
MiniMax/MiniMax-M2.1	500	20,000,000

MiMo-Xiaomi direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
xiaomi/mimo-v2.6-pro-ultraspeed	100	5,000,000
xiaomi/mimo-v2.5-pro	100	10,000,000

Stepfun-StepFun direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
stepfun/step-3.7-flash	500	20,000,000
stepfun/step-5-preview	500	10,000,000

Unisound direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
unisound/unisound-u2	300	3,000,000

Ling-AntDigital direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
antdigital/Ling-3.1-flash	300	10,000,000

Image generation

Qwen-Image

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Task submission API call limit	Number of concurrent processing tasks (concurrency)
qwen-image-3.0-pro	5 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-3.0	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-pro	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-turbo	120 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.0-pro	2 requests/minute	No limit for sync APIs
qwen-image-2.0-pro-2026-06-22	2 requests/minute	No limit for sync APIs
qwen-image-2.0-pro-2026-04-22	2 requests/minute	No limit for sync APIs
qwen-image-2.0-pro-2026-03-03	2 requests/minute	No limit for sync APIs
qwen-image-2.0	2 requests/second	No limit for sync APIs
qwen-image-2.0-2026-03-03	2 requests/second	No limit for sync APIs
qwen-mt-image-2.0	60 requests/minute	No limit for sync APIs / 2 for async APIs
qwen-mt-image	1 request/second	No limit for sync APIs / 2 for async APIs

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Task submission API call limit	Number of concurrent processing tasks (concurrency)
qwen-image-3.0-pro	Global	5 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-3.0	Global	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-pro	Global	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-turbo	Global	120 request/minute	No limit for sync APIs / 10 for async APIs

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Task submission API call limit	Number of concurrent processing tasks (concurrency)
qwen-image-3.0-pro	International	5 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-3.0	International	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-pro	International	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-turbo	International	120 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.0-pro	International	2 times per minute	No limit for sync APIs
qwen-image-2.0-pro-2026-06-22	International	2 times per minute	No limit for sync APIs
qwen-image-2.0-pro-2026-04-22	International	2 times per minute	No limit for sync APIs
qwen-image-2.0-pro-2026-03-03	International	2 requests/minute	No limit for sync APIs
qwen-image-2.0	International	2 requests/second	No limit for sync APIs
qwen-image-2.0-2026-03-03	International	2 requests/second	No limit for sync APIs
qwen-mt-image-2.0	International	60 requests/minute	No limit for sync APIs / 2 for async APIs

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Task submission API call limit	Number of concurrent processing tasks (concurrency)
qwen-image-3.0-pro	Global	5 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-3.0	Global	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-pro	Global	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-turbo	Global	120 request/minute	No limit for sync APIs / 10 for async APIs

Japan (Tokyo)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Task submission API call limit	Number of concurrent processing tasks (concurrency)
qwen-image-3.0-pro	Global	5 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-3.0	Global	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-pro	Global	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-turbo	Global	120 request/minute	No limit for sync APIs / 10 for async APIs

Hong Kong (China)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Task submission API call limit	Number of concurrent processing tasks (concurrency)
qwen-image-3.0-pro	Global	5 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-3.0	Global	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-pro	Global	20 request/minute	No limit for sync APIs / 10 for async APIs
qwen-image-2.1-turbo	Global	120 request/minute	No limit for sync APIs / 10 for async APIs

Text-to-image-Z-Image

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
z-image-turbo	2	No limit for sync APIs

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
z-image-turbo	International	2	No limit for sync APIs

Wanxiang

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
wan2.7-image-pro	5	5
wan2.7-image	5	5
wan2.6-image	5	5
wan2.6-t2i	1	5
wan2.5-t2i-preview	5	5
wan2.2-t2i-plus	2	2
wan2.2-t2i-flash	2	2
wanx2.1-t2i-plus	2	2
wanx2.1-t2i-turbo	2	2
wanx2.0-t2i-turbo	2	2
wan2.5-i2i-preview	5	5
wanx2.1-imageedit	2	2
wanx-v1	2	1
wanx-x-painting	2	1
wanx-sketch-to-image-lite	2	1

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
wan2.6-t2i	Global	5	5
wan2.6-image	Global	5	5

Singapore

Model name	Service deployment scope	Rate limiting conditions (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent tasks
wan2.7-image-pro	International	5	5
wan2.7-image	International	5	5
wan2.6-image	International	5	5
wan2.6-t2i	International	5	5
wan2.5-t2i-preview	International	5	5
wan2.2-t2i-flash	International	2	2
wan2.2-t2i-plus	International	2	2
wan2.1-t2i-turbo	International	2	2
wan2.1-t2i-plus	International	2	2
wan2.5-i2i-preview	International	5	5

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
wan2.6-t2i	Global	1	5
wan2.6-image	Global	5	5

Image Editing and Generation

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
shoemodel-v1	2	1
wanx-virtualmodel	2	1
wanx-style-repaint-v1	2	2
wanx-poster-generation-v1	2	1
virtualmodel-v2	2	1
image-instance-segmentation	2	1
image-erase-completion	2	1

FaceChain

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent tasks
facechain-facedetect	5	No limit for sync APIs
facechain-finetune	1	1

Kling series

China (Beijing)

Model name

Rate limits (triggered if any value is exceeded)

Requests per second (RPS)

Number of concurrent processing tasks (concurrency)

kling/kling-v3-omni-image-generation

For a single Alibaba Cloud Model Studio API key, the five models in the Kling series (image and video) share a concurrency limit of 10. This means that the total number of running tasks for these five models cannot exceed 10.

kling/kling-v3-image-generation

Vidu series

China (Beijing)

Model name	Rate limiting conditions (throttling is triggered when any value is exceeded)
Model name	RPM (requests per minute)	Number of concurrent tasks
vidu/vidu-image_reference2image	300	5 For a single Alibaba Cloud Model Studio API key, the Vidu reference image generation models share a concurrency limit of 5. This means that the total number of running tasks for these models cannot exceed 5.
vidu/vidu-image-pro_reference2image
vidu/vidu-image-lite_reference2image
vidu/viduq3-fast_reference2image
vidu/viduq2-pro_reference2image
vidu/viduq2-fast_reference2image

Speech recognition

Qwen-Audio-ASR

China (Beijing)

Model name	Requests per minute (RPM)
qwen-audio-3.1-asr-flash	600
qwen-audio-3.0-asr-flash	600

Singapore

Model name	Service deployment scope	Requests per minute (RPM)
qwen-audio-3.1-asr-flash	International	600
qwen-audio-3.0-asr-flash	International	600

Qwen-Audio-ASR-Streaming

China (Beijing)

Model name	Requests per minute (RPM)
qwen-audio-3.1-asr-flash-streaming	1200
qwen-audio-3.0-asr-flash-streaming	1200

Singapore

Model name	Service deployment scope	Requests per minute (RPM)
qwen-audio-3.1-asr-flash-streaming	International	1200
qwen-audio-3.0-asr-flash-streaming	International	1200

Qwen-Audio-ASR-Filetrans

China (Beijing)

Model name	Requests per minute (RPM)
qwen-audio-3.1-asr-flash-filetrans	600
qwen-audio-3.0-asr-flash-filetrans	600

Singapore

Model name	Service deployment scope	Requests per minute (RPM)
qwen-audio-3.1-asr-flash-filetrans	International	600
qwen-audio-3.0-asr-flash-filetrans	International	600

Qwen-Audio-ASR-Message

China (Beijing)

Model name	Requests per minute (RPM)
qwen-audio-3.1-asr-flash-message	1200

Singapore

Model name	Service deployment scope	Requests per minute (RPM)
qwen-audio-3.1-asr-flash-message	International	1200

Qwen3-ASR

China (Beijing)

Qwen3-ASR-Flash-Filetrans

Model name	Requests per minute (RPM)
qwen3-asr-flash-filetrans	100

Qwen3-ASR-Flash

Model name	Requests per minute (RPM)
qwen3-asr-flash	100

Singapore

Qwen3-ASR-Flash-Filetrans

Model name	Service deployment scope	Requests per minute (RPM)
qwen3-asr-flash-filetrans	International	100

Qwen3-ASR-Flash

Model name	Service deployment scope	Requests per minute (RPM)
qwen3-asr-flash	International	100

US (Virginia)

Model name	Service deployment scope	Requests per minute (RPM)
qwen3-asr-flash	United States	100

Qwen3-ASR-Realtime

China (Beijing)

Model name	Requests per second (RPS)
qwen3-asr-flash-realtime	20

Singapore

Model name	Service deployment scope	Requests per second (RPS)
qwen3-asr-flash-realtime	International	20

Fun-ASR

China (Beijing)

Model name	Requests per minute (RPM)
fun-asr	600
fun-asr-flash-2026-06-15

Singapore

Model name	Service deployment scope	Requests per minute (RPM)
fun-asr	International	600
fun-asr-flash-2026-06-15	International	600

Fun-ASR-Realtime

China (Beijing)

Model name	Requests per second (RPS)
fun-asr-realtime	20
fun-asr-realtime-2026-02-28
fun-asr-flash-8k-realtime
fun-asr-flash-8k-realtime-2026-01-28

Singapore

Model name	Service deployment scope	Requests per second (RPS)
fun-asr-realtime	International	20

Paraformer

China (Beijing)

Model name	Requests per second (RPS)
paraformer-realtime-v2	20
paraformer-realtime-8k-v2

Model name	Requests per minute (RPM)
paraformer-v2	1,200

Model name	Requests per second (RPS)	Number of tasks being processed simultaneously (concurrency)
paraformer-8k-v2	20	100

Speech synthesis

Qwen-Audio-TTS

China (Beijing)

Model name	Requests per minute (RPM)
qwen-audio-3.1-tts-plus	180
qwen-audio-3.1-tts-flash	180
qwen-audio-3.0-tts-plus	180
qwen-audio-3.0-tts-flash	180

Singapore

Model name	Service deployment scope	Requests per minute (RPM)
qwen-audio-3.1-tts-plus	International	180
qwen-audio-3.1-tts-flash	International	180
qwen-audio-3.0-tts-plus	International	180
qwen-audio-3.0-tts-flash	International	180

CosyVoice

China (Beijing)

Model name	Requests per second (RPS)
cosyvoice-v3.5-plus	3
cosyvoice-v3.5-flash
cosyvoice-v3-plus
cosyvoice-v3-flash
cosyvoice-v2

Singapore

Model name	Service deployment scope	Requests per second (RPS)
cosyvoice-v3-plus	International	3
cosyvoice-v3-flash	International	3

MiniMax-Xiyu Tech direct supply

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
MiniMax/speech-2.8-hd	20	20,000
MiniMax/speech-02-hd	20	20,000
MiniMax/speech-2.8-turbo	20	20,000
MiniMax/speech-02-turbo	20	20,000

Qwen3-TTS

China (Beijing)

Qwen3-TTS-Flash

Model name	Requests per minute (RPM)
qwen3-tts-flash	180

Singapore

Qwen3-TTS-Flash

Model name	Service deployment scope	Requests per minute (RPM)
qwen3-tts-flash	International	180

Qwen3-TTS-Realtime

China (Beijing)

Qwen3-TTS-Flash-Realtime

Model name	Requests per minute (RPM)
qwen3-tts-flash-realtime	180

Singapore

Qwen3-TTS-Flash-Realtime

Model name	Service deployment scope	Requests per minute (RPM)
qwen3-tts-flash-realtime	International	180

Sambert

China (Beijing)

Model service	Requests per second (RPS)
Sambert series models	20

Audio generation

Qwen-Audio-TTS-Next

China (Beijing)

Model	Requests per minute (RPM)
qwen-audio-3.1-tts-next	180

Singapore

Model	Requests per minute (RPM)
qwen-audio-3.1-tts-next	180

Voice chat

Qwen-Audio-Realtime

China (Beijing)

Model name	Rate limiting conditions (triggered if any value is exceeded) The following limits are per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen-audio-3.1-realtime-plus	60	100,000
qwen-audio-3.1-realtime-flash	60	100,000
qwen-audio-3.0-realtime-plus	60	100,000
qwen-audio-3.0-realtime-flash	60	100,000

Singapore

Model name	Service deployment scope	Rate limiting conditions (triggered if any value is exceeded) The following limits are per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Including input and output tokens
qwen-audio-3.1-realtime-plus	International	60	100,000
qwen-audio-3.1-realtime-flash	International	60	100,000
qwen-audio-3.0-realtime-plus	International	60	100,000
qwen-audio-3.0-realtime-flash	International	60	100,000

Audio and video translation

Qwen3-LiveTranslate-Flash

China (Beijing)

Model name	Rate limiting conditions (rate limiting is triggered when any value is exceeded) The following are per-minute rate limiting conditions. The service may also enforce RPS (RPM/60) and TPS (TPM/60) limits
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Including input and output tokens
qwen3-livetranslate-flash	100	100,000
qwen3-livetranslate-flash-2025-12-01	100	100,000

Singapore

Model name	Service deployment scope	Rate limiting conditions (rate limiting is triggered when any value is exceeded) The following are per-minute rate limiting conditions. The service may also enforce RPS (RPM/60) and TPS (TPM/60) limits
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Including input and output tokens
qwen3-livetranslate-flash	International	100	100,000
qwen3-livetranslate-flash-2025-12-01	International	100	100,000

Qwen-LiveTranslate-Flash-Realtime

China (Beijing)

Model name	Rate limiting conditions (rate limiting is triggered when any value is exceeded) The following are per-minute rate limiting conditions. The service may also enforce RPS (RPM/60) and TPS (TPM/60) limits
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Including input and output tokens
qwen3.8-livetranslate-flash-realtime	10	100,000
qwen3.5-livetranslate-flash-realtime
qwen3.5-livetranslate-flash-realtime-2026-05-19

Singapore

Model name	Service deployment scope	Rate limiting conditions (rate limiting is triggered when any value is exceeded) The following are per-minute rate limiting conditions. The service may also enforce RPS (RPM/60) and TPS (TPM/60) limits
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Including input and output tokens
qwen3.8-livetranslate-flash-realtime	International	10	100,000
qwen3.5-livetranslate-flash-realtime	International
qwen3.5-livetranslate-flash-realtime-2026-05-19	International

Voice cloning and design

Qwen-Audio-TTS/CosyVoice voice cloning/design

Qwen-Audio-TTS/CosyVoice voice cloning/design models share a single model and a single rate limit quota.

China (Beijing)

Model name	Requests per second (RPS)
voice-enrollment	10

Singapore

Model name	Service deployment scope	Requests per second (RPS)
voice-enrollment	International	10

Qwen3-TTS voice cloning

China (Beijing)

Model name	Requests per minute (RPM)
qwen-voice-enrollment	180

Singapore

Model name	Service deployment scope	Requests per minute (RPM)
qwen-voice-enrollment	International	180

Music generation

Fun-Music

China (Beijing)

Model name	Requests per minute (RPM)
fun-music-preview	180
fun-music-v1	180

Video generation

HappyHorse series

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
happyhorse-1.1-t2v	5	5
happyhorse-1.1-i2v	5	5
happyhorse-1.1-r2v	5	5
happyhorse-1.0-t2v	5	5
happyhorse-1.0-i2v	5	5
happyhorse-1.0-r2v	5	5
happyhorse-1.0-video-edit	5	5

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
happyhorse-1.1-t2v	Global	5	5
happyhorse-1.1-i2v	Global	5	5
happyhorse-1.1-r2v	Global	5	5
happyhorse-1.0-t2v	Global	5	5
happyhorse-1.0-i2v	Global	5	5
happyhorse-1.0-r2v	Global	5	5
happyhorse-1.0-video-edit	Global	5	5

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
happyhorse-1.1-t2v	International	5	5
happyhorse-1.1-i2v	International	5	5
happyhorse-1.1-r2v	International	5	5
happyhorse-1.0-t2v	International	5	5
happyhorse-1.0-i2v	International	5	5
happyhorse-1.0-r2v	International	5	5
happyhorse-1.0-video-edit	International	5	5

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
happyhorse-1.1-t2v	Global	5	5
happyhorse-1.1-i2v	Global	5	5
happyhorse-1.1-r2v	Global	5	5
happyhorse-1.0-t2v	Global	5	5
happyhorse-1.0-i2v	Global	5	5
happyhorse-1.0-r2v	Global	5	5
happyhorse-1.0-video-edit	Global	5	5

Japan (Tokyo)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
happyhorse-1.1-t2v	Global	5	5
happyhorse-1.1-i2v	Global	5	5
happyhorse-1.1-r2v	Global	5	5
happyhorse-1.0-video-edit	Global	5	5

Wanxiang series

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
wan3.0-video-prime	5	5
wan3.0-video	5	5
wan2.7-r2v-2026-06-12	5	5
wan2.7-t2v-2026-06-12	5	5
wan2.7-t2v-2026-04-25	5	5
wan2.7-t2v	5	5
wan2.6-t2v	5	5
wan2.5-t2v-preview	5	5
wan2.2-t2v-plus	2	2
wanx2.1-t2v-turbo	2	2
wanx2.1-t2v-plus	2	2
wan2.7-i2v-2026-04-25	5	5
wan2.7-i2v	5	5
wan2.6-i2v-flash	5	5
wan2.6-i2v	5	5
wan2.5-i2v-preview	5	5
wan2.2-i2v-flash	2	2
wan2.2-i2v-plus	2	2
wanx2.1-i2v-turbo	2	2
wanx2.1-i2v-plus	2	2
wan2.2-kf2v-flash	2	2
wanx2.1-kf2v-plus	2	2
wanx2.1-vace-plus	2	2
wan2.7-videoedit	5	5
wan2.7-r2v	5	5
wan2.6-r2v-flash	5	5
wan2.6-r2v	5	5
wan2.2-s2v-detect	5	No limit for sync APIs
wan2.2-s2v	5	1
wan2.2-animate-move	5	1
wan2.2-animate-mix	5	1

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
wan3.0-video-prime	Global	5	5
wan3.0-video	Global	5	5
wan2.6-t2v	Global	5	5
wan2.6-i2v	Global	5	5
wan2.6-r2v	Global	5	5
wan2.6-t2v	US	5	5
wan2.6-i2v	US	5	5

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
wan3.0-video-prime	International	5	5
wan3.0-video	International	5	5
wan2.7-r2v-2026-06-12	Global	5	5
wan2.7-t2v-2026-06-12	International	5	5
wan2.7-t2v-2026-04-25	International	5	5
wan2.7-t2v	International	5	5
wan2.6-t2v	International	5	5
wan2.5-t2v-preview	International	5	5
wan2.2-t2v-plus	International	2	2
wan2.1-t2v-turbo	International	2	2
wan2.1-t2v-plus	International	2	2
wan2.7-i2v-2026-04-25	International	5	5
wan2.7-i2v	International	5	5
wan2.6-i2v-flash	International	5	5
wan2.6-i2v	International	5	5
wan2.5-i2v-preview	International	5	5
wan2.2-i2v-plus	International	2	2
wan2.1-i2v-turbo	International	2	2
wan2.1-i2v-plus	International	2	2
wan2.1-kf2v-plus	International	1	2
wan2.1-vace-plus	International	2	2
wan2.7-videoedit	International	5	5
wan2.7-r2v	International	5	5
wan2.6-r2v-flash	International	5	5
wan2.6-r2v	International	5	5
wan2.2-animate-move	International	5	1
wan2.2-animate-mix	International	5	1

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded)
Model name	Service deployment scope	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
wan3.0-video-prime	Global	5	5
wan3.0-video	Global	5	5
wan2.6-t2v	Global	5	5
wan2.6-i2v	Global	5	5
wan2.6-r2v	Global	5	5

Japan (Tokyo)

Model name	Service region	Rate limiting conditions (triggered when any value is exceeded)
Model name	Service region	Requests per second (RPS)	Number of concurrent processing tasks
wan3.0-video-prime	Global	5	5
wan3.0-video	Global	5	5

China (Hong Kong)

Model name	Service region	Rate limiting conditions (triggered when any value is exceeded)
Model name	Service region	Requests per second (RPS)	Number of concurrent processing tasks
wan3.0-video-prime	Global	5	5
wan3.0-video	Global	5	5

AnimateAnyone

China (Beijing)

Model name	Requests per second (RPS)	Number of concurrent processing tasks
animate-anyone-detect	5	1 computing unit supports 2 concurrent tasks
animate-anyone	5	1 computing unit supports 1 concurrent task

VideoRetalk

China (Beijing)

Model name

Requests per second (RPS)

Number of concurrent processing tasks

videoretalk

At any given time, only one job is in the running state. Other jobs in the queue are in a waiting state.

Video Style Transfer

China (Beijing)

Model name

Requests per second (RPS)

Number of concurrent processing tasks

video-style-transform

At any given time, only one job is in the running state. Other jobs in the queue are in a waiting state.

Aishell series

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Task submission API RPM limit	Number of concurrent processing tasks (concurrency)
pixverse/pixverse-lipsync	300	5 The quota is shared among the 3 models under the same Model Studio API key. The total number of running tasks for these 3 models cannot exceed 5.
pixverse/pixverse-motioncontrol	300
pixverse/pixverse-upscale	300

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
pixverse/pixverse-c1-t2v	5	5 The quota is shared among the four models under the same Model Studio API key. The total number of running tasks for these four models cannot exceed 5.
pixverse/pixverse-c1-it2v
pixverse/pixverse-c1-kf2v
pixverse/pixverse-c1-r2v
pixverse/pixverse-v6-r2v-omni	5	5 The quota is shared among the five models under the same Model Studio API key. The total number of running tasks for these five models cannot exceed 5.
pixverse/pixverse-v6-t2v
pixverse/pixverse-v6-it2v
pixverse/pixverse-v6-kf2v
pixverse/pixverse-v6-r2v
pixverse/pixverse-v5.6-t2v	5	5 The quota is shared among the four models under the same Model Studio API key. The total number of running tasks for these four models cannot exceed 5.
pixverse/pixverse-v5.6-it2v
pixverse/pixverse-v5.6-kf2v
pixverse/pixverse-v5.6-r2v

Kling series

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
kling/kling-v3-omni-video-generation	5	10 The five models in the Kling series (image and video) share a concurrency of 10 under the same Model Studio API key. This means that the total number of running tasks for these five models cannot exceed 10.
kling/kling-v3-turbo-video-generation
kling/kling-v3-video-generation

Vidu series

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per second (RPS)	Number of concurrent processing tasks (concurrency)
vidu/viduq3-ad_reference2video	5	5 The concurrency quota is shared among the 20 models under the same Model Studio API key. The total number of running tasks for these 20 models cannot exceed 5.
vidu/viduq3-drama_reference2video	5
vidu/viduq3-pro-fast_img2video	5
vidu/viduq3-turbo_text2video	5
vidu/viduq3-pro_text2video	5
vidu/viduq2_text2video	5
vidu/viduq3-turbo_img2video	5
vidu/viduq3-pro_img2video	5
vidu/viduq2-turbo_img2video	5
vidu/viduq2-pro_img2video	5
vidu/viduq2-pro-fast_img2video	5
vidu/viduq3-turbo_start-end2video	5
vidu/viduq3-pro_start-end2video	5
vidu/viduq2-turbo_start-end2video	5
vidu/viduq2-pro_start-end2video	5
vidu/viduq3-mix_reference2video	5
vidu/viduq3_reference2video	5
vidu/viduq3-turbo_reference2video	5
vidu/viduq2-pro_reference2video	5
vidu/viduq2_reference2video	5

MiniMax series

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Task submission API RPM limit	Maximum concurrent tasks
MiniMax/MiniMax-H3	300	5

3D model generation

Tripo series

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Task submission API RPM limit	Number of concurrent processing tasks (concurrency)
Tripo/Tripo-H3.1	300	10
Tripo/Tripo-P1.0	300	5
Tripo/Tripo-ImageToSplat	300	5
Tripo/Tripo-AutoRig	300	10 Auto rigging, rig check and animation retargeting share the same concurrency quota. The total number of running tasks for these three models cannot exceed 10.
Tripo/Tripo-RigCheck	300
Tripo/Tripo-AnimationRetarget	300
Tripo/Tripo-Texture	300	5 Texturing and format conversion share the same concurrency quota. The total number of running tasks for these two models cannot exceed 5.
Tripo/Tripo-ConvertFormat	300
Tripo/Tripo-Segmentation	300	10 Segmentation, smart segmentation, mesh completion and retopology share the same concurrency quota. The total number of running tasks for these four models cannot exceed 10.
Tripo/Tripo-SmartSegmentation	300
Tripo/Tripo-MeshCompletion	300
Tripo/Tripo-Retopology	300

Embedding and sorting models

Text embedding

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded)
Model name	Requests per minute (RPM)	Tokens per minute (TPM)/Jobs Input tokens only
qwen3.7-text-embedding	24,000	1,000,000
qwen3.7-text-embedding-flash	24,000	1,000,000
text-embedding-v1 Rate limiting does not apply when calling the service with the Batch API.	1,800	1,200,000
text-embedding-v2 Rate limiting does not apply when calling the service with the Batch API.	1,800	1,200,000
text-embedding-v3 Rate limiting does not apply when calling the service with the Batch API.	1,800	1,200,000
text-embedding-v4 Rate limiting does not apply when calling the service with the Batch API.	1,800	1,200,000
text-embedding-async-v1	60	The number of queued and running asynchronous text embedding jobs for a user cannot exceed 50. To prevent resource contention from burst jobs, the number of concurrent jobs is limited to 3. This means a single user can have a maximum of 3 asynchronous text embedding jobs running at the same time. Other jobs will be queued.
text-embedding-async-v2	60	The number of queued and running asynchronous text embedding jobs for a user cannot exceed 50. To prevent resource contention from burst jobs, the number of concurrent jobs is limited to 3. This means a single user can have a maximum of 3 asynchronous text embedding jobs running at the same time. Other jobs will be queued.

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.7-text-embedding	International	24,000	1,000,000
text-embedding-v4	International	1,800	1,000,000
text-embedding-v3	International	6,000	24,000,000

Hong Kong (China)

Model name	Service deployment scope	Rate limiting conditions (triggered if any value is exceeded) The following limits are per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens.
text-embedding-v4	Hong Kong (China)	1,800	1,200,000

Multimodal embedding

China (Beijing)

Model name	Rate Limiting Conditions The following are the limits per minute. The service may also enforce limits based on requests per second (RPS = RPM/60) and tokens per second (TPS = TPM/60).
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Input tokens only
qwen3-vl-embedding	2,400	1,200,000
qwen2.5-vl-embedding	1,200	600,000
tongyi-embedding-vision-plus	600	200,000
tongyi-embedding-vision-flash	600	200,000
tongyi-embedding-vision-flash-2026-03-06	1,200	9,600,000
tongyi-embedding-vision-plus-2026-03-06	1,200	9,600,000
multimodal-embedding-v1	120	1,000,000

Reranking Models

China (Beijing)

Model name	Rate limiting conditions (triggered when any value is exceeded) The following are the per-minute rate limits. The service may also enforce limits based on requests per second (RPS), which is RPM/60, and tokens per second (TPS), which is TPM/60.
Model name	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3.7-text-rerank	3,000	9,000,000
qwen3-rerank	5,400	5,000,000,000
qwen3-vl-rerank	600	9,000,000
gte-rerank-v2	5,040	4,980,000,000

Singapore

Model name	Service deployment scope	Rate limiting conditions (triggered when any value is exceeded) The following are the per-minute rate limits. The service may also enforce limits based on requests per second (RPS), which is RPM/60, and tokens per second (TPS), which is TPM/60.
Model name	Service deployment scope	Requests per minute (RPM)	Tokens per minute (TPM) Includes input and output tokens
qwen3-rerank	International	5,400	5,000,000,000
gte-rerank-v2	International	5,040	4,980,000,000

Industry

Intention recognition

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
tongyi-intent-detect-v3	1,200	1,000,000

Role assumption

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
qwen-plus-character	120	500,000
qwen-flash-character	120	500,000
qwen-flash-character-2026-02-26	120	500,000

Singapore

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Service deployment scope	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
qwen-plus-character	International	120	500,000
qwen-flash-character	International	120	500,000
qwen-plus-character-ja	International	120	500,000

US (Virginia)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Service deployment scope	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
qwen-plus-character	Global	120	500,000
qwen-flash-character	Global	120	500,000

Germany (Frankfurt)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Service deployment scope	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
qwen-plus-character	Global	120	5,000,000

Japan (Tokyo)

Model name	Service deployment scope	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Service deployment scope	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
qwen-plus-character	Global	120	5,000,000

Interface interaction

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
gui-plus	80	540,000
gui-plus-2026-02-26	100	540,000

Decision model

China (Beijing)

Model name	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
decision-model-preview	1,200	2,000,000

Singapore

Model name	Rate limits (triggered if any value is exceeded) The limits below are specified per minute. The service may enforce limits based on records per second (RPS) (RPM/60) and Transactions Per Second (TPS) (TPM/60).
Model name	Requests Per Minute (RPM)	Tokens Per Minute (TPM) Includes input and output tokens
decision-model-preview	1,200	2,000,000

Unpublished models

For more information, see Model unpublishing policy.

FAQ

Why is rate limiting triggered?

You can identify the type of rate limit triggered based on the error message:

Requests rate limit exceeded or You exceeded your current requests list: This indicates that the requests per minute (RPM) limit was triggered.
Allocated quota exceeded or You exceeded your current quota: This indicates that the tokens per minute (TPM) limit was triggered.
Request rate increased too quickly: The request frequency surged in a short period, triggering system stability protection. This can occur even if the total number of calls has not reached the RPM or TPM limits.
For other errors, see Error codes to confirm the cause.

A burst of requests in a short period can trigger rate limiting even if the per-minute limits are not exceeded. For the per-second enforcement details, see Static rate limiting rules.

How to view model usage

You can view the data in the monitoring charts within minutes after calling a model. Log in to the Model Monitoring page (Beijing or Singapore), select O&M Management > Model Monitoring in the left navigation to enter the monitoring overview page. Set the query conditions (such as time range and workspace). Then, in the Models area, find the target model and click View Details to view the model's call statistics. For more information, see the Monitoring and Alert document.

Monitoring data is updated at minute-level and is for reference only, not for billing.

How long does it take to recover from rate limiting?

Recovery usually occurs within one minute. If other errors occur, see Error codes for troubleshooting.

Model response speed is not related to whether you use free quota or paid calls. Free quota and paid calls use the same model service infrastructure, so the generation speed for a given model is the same. Free quota only controls whether a request is accepted: once the free quota is exhausted, the request is rejected with a 403 error (if you have enabled stop when the free quota is used up). This does not reduce the generation speed of any request that has already been accepted.

What factors affect model response speed?

Model type: lightweight models (such as qwen-flash) generate responses faster than larger models (such as qwen-max).
Output length: the more tokens a response contains, the longer the total generation time.
Server load: response speed may fluctuate slightly during peak periods.

Does rate limiting reduce the generation speed of accepted requests?

No. Rate limiting (RPM/TPM) only rejects requests that exceed the limit, returning a 429 error. When the free quota is exhausted (if you have enabled stop when the free quota is used up), requests are rejected with a 403 error. Neither case affects the generation speed of requests that have already been accepted: 429 indicates rate limiting, and 403 indicates quota exhaustion — both are rejection mechanisms, not speed throttling.

How to avoid rate limiting

Choose models with higher rate limits
- Prioritize models with higher rate limits, such as qwen-plus.
- Stable or latest versions have more lenient rate limits than dated snapshot versions.
Optimize your call strategy
- Reduce call frequency: If you receive a Requests rate limit exceeded or You exceeded your current requests list error, lower the API call frequency.
- Reduce token consumption: If you receive an Allocated quota exceeded or You exceeded your current quota error, shorten the input or limit the output length.
- Smooth the request rate: If you receive a Request rate increased too quickly error, use uniform scheduling, exponential backoff, or a request queue to distribute requests evenly and avoid sudden peaks.

Add a backup model

If rate limiting is triggered, you can switch to a backup model to continue generation. This can reduce the probability of failure and increase throughput. The following code automatically retries with qwen-flash after a rate limit is triggered for qwen-plus. You can also choose a model from a different series (such as qwen-flash) as the backup model to further reduce the risk that the primary and backup models are rate limited at the same time.

Sample code

import os
import asyncio
from openai import AsyncOpenAI, APIStatusError

# Configure
API_KEY = os.getenv("DASHSCOPE_API_KEY")
# Primary model
MODEL = "qwen-plus"
# Backup model
BACKUP_MODEL = "qwen-flash"
# Test question
QUESTION = "Who are you?"
# Concurrency setting
NUM_REQUESTS = 10

client = AsyncOpenAI(
    api_key=API_KEY,
    # The following is the URL for the China (Beijing) region. Replace {WorkspaceId} with your actual workspace ID.
    base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1"
)

async def send_request(model):
    """Sends a single request."""
    try:
        await client.chat.completions.create(
            model=model,
            messages=[{"role": "user", "content": QUESTION}]
        )
        return True
    except APIStatusError as e:
        if e.status_code == 429:
            print(f"[Rate limit triggered] Model {model}")
            return False
        raise
    except Exception as e:
        print(f"[Request failed] Model {model}, Error: {e}")
        return False

async def task(i):
    # Try the primary model.
    if await send_request(MODEL):
        return True
    # If rate limited, try the backup model.
    return await send_request(BACKUP_MODEL)

async def main():
    results = await asyncio.gather(*(task(i) for i in range(NUM_REQUESTS)))
    print(f"Successful requests: {sum(results)}, Failed requests: {len(results) - sum(results)}")

if __name__ == "__main__":
    asyncio.run(main())

Split tasks: Long conversations or large documents can consume many tokens quickly. You can split large batch tasks into smaller batches and submit them at different times.
Use batch inference: For tasks that do not require real-time responses, you can use the Batch API. Batch requests are not subject to real-time rate limits, but you must consider queuing and processing time.

How to control token usage or costs

Rate limiting only restricts the request rate per unit of time; it does not cap cumulative usage. To control token usage or costs, use the following methods:

Set a spending limit and cost alerts: On the Billing card, configure Cost alerts to enable a monthly spending limit and threshold notifications. You are notified when the threshold is reached, which helps you avoid overspending. For more information, see Query bills and manage costs.
Enable stop when the free quota is used up: For models that offer a free quota, you can enable stop when the free quota is used up so that calls stop automatically once the free quota is exhausted, which prevents additional charges. For more information, see Free quota.
Monitor model usage: Regularly check the token usage of each model to detect abnormal growth in time. See How to view model usage above.

Will I still be rate limited after I top up my account?

Topping up your account does not change the RPM and TPM rate limits of models that use static rate limiting. Rate limits are configured at the Alibaba Cloud account level and are independent of billing. Topping up (pay-as-you-go) only ensures that your account is not suspended for overdue payments; it does not raise static rate limits. For models that use dynamic rate limiting, the TPM rate limit tier is automatically adjusted based on your monthly consumption amount, and the tier can be upgraded as consumption increases. For the adjustment rules, see Dynamic rate limit adjustment strategy.

To obtain higher rate limits, contact your account manager.